Direct extrapolation of survey results on purchase intentions may give a biased view on actual consumer behavior. This is because the purchase intentions of consumers may be affected by the survey itself. On the positive side, such effects can be incorporated in econometric models to get reliable estimates of actual behavior of non-surveyed consumers, which often is the ultimate purpose of survey studies. This paper proposes a reasonably simple methodology to correct for such possible survey effects and to get consistent pre- dictions beyond the survey sample. The potential merits of the method are illustrated by a supermarket survey on easy-to-prepare food products and related health issues. This indicates that the required corrections can be quite substantial and that predictions that neglect survey effects can be seriously biased indeed.

bias correction, consumer behavior, econometric models, purchase prediction, survey effects
Econometric Institute Research Papers
Report / Econometric Institute, Erasmus University Rotterdam
Erasmus School of Economics

Heij, C, & Franses, Ph.H.B.F. (2006). Prediction beyond the survey sample: correcting for survey effects on consumer decisions. (No. EI 2006-48). Report / Econometric Institute, Erasmus University Rotterdam. Retrieved from